mms-1b-all-bemgen-combined-train-drodat-gdro-3e-2-dat-5e-1-no-sd-0.00-42
This model is a fine-tuned version of facebook/mms-1b-all on the BEMGEN - BEM dataset. It achieves the following results on the evaluation set:
- Loss: 12.6062
- Cer: 0.0541
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 8
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 278.7486 | 1.2690 | 500 | 236.9604 | 0.9669 |
| 36.4145 | 2.5381 | 1000 | 14.6363 | 0.0633 |
| 31.2453 | 3.8071 | 1500 | 13.8659 | 0.0611 |
| 30.9025 | 5.0761 | 2000 | 13.7702 | 0.0595 |
| 30.5784 | 6.3452 | 2500 | 13.4068 | 0.0579 |
| 28.1045 | 7.6142 | 3000 | 13.1432 | 0.0568 |
| 25.8265 | 8.8832 | 3500 | 13.2758 | 0.0576 |
| 27.6642 | 10.1523 | 4000 | 13.0062 | 0.0563 |
| 25.0027 | 11.4213 | 4500 | 13.0657 | 0.0570 |
| 25.6513 | 12.6904 | 5000 | 13.1733 | 0.0557 |
| 23.6991 | 13.9594 | 5500 | 12.9816 | 0.0553 |
| 23.0126 | 15.2284 | 6000 | 13.0489 | 0.0550 |
| 23.8612 | 16.4975 | 6500 | 13.0871 | 0.0547 |
| 24.7649 | 17.7665 | 7000 | 12.8327 | 0.0546 |
| 23.6305 | 19.0355 | 7500 | 12.9027 | 0.0541 |
| 21.1205 | 20.3046 | 8000 | 12.6059 | 0.0541 |
| 20.4796 | 21.5736 | 8500 | 12.6218 | 0.0536 |
| 21.1323 | 22.8426 | 9000 | 12.8034 | 0.0538 |
| 19.5444 | 24.1117 | 9500 | 12.7840 | 0.0536 |
Framework versions
- Transformers 4.53.0
- Pytorch 2.10.0+cu128
- Datasets 4.5.0
- Tokenizers 0.21.4